Linear post regulator for SMPS, what's your take?

[wwenze]

I was wondering if anybody tried or thought about the idea of using a cheap SMPS compared to mains frequency transformer to get the voltage required, then use a linear regulator to clean up the noise?

Did a few simple sims and it seems pretty easy to get ripple and noise down to way below 10mV even with the worst of power supplies, regulation's pretty good too.

Efficiency's low, wide input voltage range and PFC, and little efficiency penalty if the output also needs to be adjustable using the SMPS to drop most of the voltage. With cost slightly higher than the SMPS alone and noise level somewhere between linear and switching, how sweet is that?

[AndrewT]

Originally Posted by wwenze

.....................easy to get ripple and noise down to way below 10mV even with the worst of power supplies, regulation's pretty good too.

....................Efficiency's low,

set your ripple target much lower than 10mVac, or is that 10mVpp?
There is no need for low efficiency.
With a fixed input voltage from the smps and a fixed output voltage from the linear then the voltage overhead of passing through the linear can be trimmed very low to maintain good output impedance for all likely current loadings.

[wwenze]

That was a very conservative estimate. I randomly threw together some values with the worst SMPS that could ever exist and what I got:



Think there's much room for improvement still. I'm green at this so it would really help if I could be given some pointers on how to choose the values.

[martin clark]

If you can spare the voltage overhead, a little RC ahead of the regulator helps a lot in stopping HF switching noise passing straight through the reg (which has very little gain at HF)

Example:
Using 3-pin regulators off-piste: part 2

[wwenze]

Thanks for your reply. At first I did an RC with some values but for a 6A load the C had to be too large. But decided to try it again after reading your reply. Settled for 0.1ohm 2200uF for 2A load, HF greatly reduced and ripple is below 300uVp-p.
This will do for now while I search for the appropriate component models.

Thanks AndrewT too for telling me to set my target much lower.

[Calvin]

Hi,

I´d not opt for an 3-pin Reg but a simple gyrator circuit instead, because of bandwidth issues. It needs to be effective in filtering out HF. Ripple rejection can be made high enough too, because the ripple frequency is high and the filtering Gate/Base- RC-circuit doesn´t need a low value of the R.

jauu
Calvin

[wwenze]

The output needs to be regulated and gyrator alone without a regulator seems to kill the regulation of the source. Anything I'm doing wrong?

[agdr]

Take a look at the schematic in the link at the bottom of that good page martin clark provided the link to:

Hear it in the one drop

The gyrator kills much of the wideband smps noise, then following it with your LT1083 can provide some regulation. You wouldn't use the follow-on op amp error amp, of course, with your amperage needs.

[agdr]

Maybe something like this, for 18V 6A, using your voltage sources. Fc = 4.8Hz (1k and 33uF) from that article.

Green = Q1 collector
Blue = LT1083 in
Red = LT1083 out

[wwenze]

Ahh so that's why my gyrator failed when I tried... didn't factor in enough voltage drop.

That is awesome!